🌍 DisasterVision-AI

AI-Powered Disaster Image Classification API

AI-powered REST API for real-time disaster image classification using Deep Learning and Machine Learning.

---

📖 Overview

DisasterVision-AI is a production-ready Computer Vision REST API that classifies disaster-related images in real time using a hybrid AI pipeline.

The system combines Deep Learning and Machine Learning models to analyze uploaded images and predict one of several disaster categories with high speed and reliability.

Designed with FastAPI, the project is lightweight, stateless, and easy to integrate into emergency response systems, monitoring dashboards, smart city platforms, and other AI-powered applications.

---

✨ Features

• 🤖 AI-powered image classification
• ⚡ Fast inference with FastAPI
• 🧠 CNN + Machine Learning models
• 📷 Single image prediction
• 📊 Probability scores for every class
• 🔄 Automatic model loading
• 🚀 Production-ready REST API
• 📦 Stateless architecture
• 🔌 Easy backend integration
• 📚 Interactive Swagger documentation

---

🧠 Supported Classes

Class	Description
🔥 Fire	Fire disasters
🌊 Flooded Areas	Flood detection
🚗 Traffic Incident	Road accidents
🏚️ Collapsed Building	Structural damage
🌳 Normal	No disaster detected

---

🏗️ System Architecture

Client
   │
   ▼
FastAPI REST API
   │
   ▼
Preprocessing
   │
   ▼
CNN Feature Extraction
   │
   ├─────────────┐
   ▼             ▼
Random Forest    SVM
        │
        ▼
Prediction
        │
        ▼
JSON Response

---

🛠 Tech Stack

AI & Machine Learning

• TensorFlow / Keras
• Scikit-learn

Backend

• FastAPI
• Uvicorn

Language

• Python

---

📂 Project Structure

DisasterVision-AI/

│
├── assets/
│   ├── banner.png
│   ├── logo.png
│   ├── predict.png
│   ├── workflow.png
│   └── swagger.png
│
├── models/
│   ├── disaster_model.keras
│   ├── svm_model.pkl
│   ├── rf_model.pkl
│   ├── scaler.pkl
│   └── label_encoder.pkl
│
├── app.py
├── model_wrapper.py
├── requirements.txt
└── README.md

---

🚀 Installation

Clone the repository

git clone https://github.com/Ahmred0000/DisasterVision-AI.git

Move into the project

cd DisasterVision-AI

Install dependencies

pip install -r requirements.txt

---

▶ Run the API

python -m uvicorn app:app --host 0.0.0.0 --port 8000

Open your browser

http://localhost:8000/docs

---

📡 API Endpoint

POST /predict

Content-Type

multipart/form-data

Field

file

---

📥 Example Request

curl -X POST "http://localhost:8000/predict" \
-F "file=@image.jpg"

---

📤 Example Response

{
  "predicted_class": "fire",
  "probabilities": [
    0.97,
   ]
}

---

⚡ Performance

Hardware	Inference Time
CPU	100–300 ms
GPU	10–50 ms

---

📸 Screenshots

Prediction Example

---

API Documentation

---

Workflow

---

🌐 Live Demo

Hugging Face Space

https://ahmed-13-ai-model.hf.space/predict

---

📚 Dataset

AIDER Disaster Dataset

https://zenodo.org/records/3888300

---

🔮 Future Improvements

• Docker Support
• Authentication
• Batch Prediction
• ONNX Export
• Kubernetes Deployment
• Explainable AI (Grad-CAM)
• Model Versioning
• Performance Monitoring

---

🤝 Contributing

Contributions are welcome!

Feel free to fork the project, improve it, and submit a Pull Request.

---

📄 License

This project is licensed under the MIT License.

---

👨‍💻 Author

Ahmed Ashraf

GitHub

https://github.com/Ahmred0000

---

⭐ If you found this project useful, consider giving it a Star!